J. Chem. Eng. Data 1987, 32,83-85
83
Partial Molar Excess Volumes at Infinite Dilution of Binary Liquid Mixtures of Nonelectrolytes James H. Cordray and Charles A. Eckert" Department of Chemical Engineering, University of Illinois, Urbana, Illinois 6 180 1
A vlbratlng-tube denslty meter was used to obtaln excess volumes ( vE) of binary llquld mlxtures of nonelectrolytes In the dilute reglon. The excess volume data were extrapolated to obtain partlal molar excess volumes at lnflnite dllutlon (V,Em) for 17 systems at 25 OC.
Introductlon Most of the attempts to d e l solution behavior have focused on the excess Gibbs energy (8') of mixing, which is the property generally considered to be of most interest for phase equilibria. The resulting expressions (e.g., Wilson, NRTL, UNIQUAC) are sufficiently accurate to correlate VLE data well, but they have been less successful in representing some aspects of LLE data, and they provide little information about the other excess properties which comprise the temperature and pressure derivatives of g E . These derivative relationships may be summarized as follows (dgE/dT),, =
(dg'/dP),, =
-SE
VE
where sE, h', and vE are the excess entropy, excess enthalpy, and excess volume of mixing. From a theoretical standpoint, examination of the partial molar derivative excess properties at infinite dilution (SF", hfm, ii;") appears to be of particular interest. At the limit of infinite dilution, solute-solute interactions disappear; thus, the values of the infinite dllution partial molar excess properties provide insight into the relationships between the solute-solvent interactions and the respective thermodynamic properties, independent of composition effects. As part of a continuing investigation into the properties of the derivatives of g', we have used a vibrating-tube denslty meter to measure the densities of binary liquid mixtures in the dilute region. From these data we obtained values of the partial molar excess volume at infinite dilution (8;") for 33 compounds In 17 binary systems.
Experlmental Sectlon Densities of dilute liquid mixtures at 25 OC were determined wtth a Mettler-Parr vibrating-tube density meter, the principle of operation of which has been presented previously ( 1 , 2). The denslty meter was calibrated daily with distilled water and air. The liquid mixtures were prepared by weight in 30-mL flasks. I n performing the weighings, a flask was first weighed with a septum cap and magnetic stirring bar. Then approximately 25 cm3 of solvent was added to the flask, the flask was capped, and the flask and solvent were weighed. Then a small amount of solute was injected through the septum into the flask, and the weighing was repeated. The vapor space in the flask was approximately 2-3 cm3. Weighings were performed on a Mettler Type H16 balance accurate to fO.1 mg, and buoyancy corrections were applied
Table I. Analysis of v E Data and Determination of PIE" for the Chloroform (1)-Methyl Acetate (2) Systemn X1
uE(exptI), cm3/mol
0.0101 0.0203 0.0309 0.0364 0.0448
0.0137 0.0242 0.0349 0.0426 0.0520
uE(calcd), cm3/mol linear fit quadratic fit* 0.0117 0.0237 0.0360 0.0424 0.0521 A = 1.16 u = 0.0013
E = 0.0020
0.0122 0.0242 0.0363 0.0424 0.0515 A = 1.23 B = -1.70 u = 0.0012 E = 0.0015
(z(uEcdd
Ou = - U~,,,~)~/(NDAT- NPAR))l/* where NDAT is the number of data points and NPAR is the number of adjustable parameters in the fit. E = maximum error of the fit. A and B are the parameters obtained from the least-squares fit of the data. *DIE" = 1.23 cm3/mol.
to account for air displacement by the liquid. After the third and final weighing, the dilute solution was mixed thoroughly and a sample was injected into the cell of the density meter, where its density was recorded. Constant temperature water flowed through a water jacket surrounding the cell, reducing temperature fluctuations in the cell to f0.003 OC during the course of a run. The absolute temperature of the cell was measured with a thermistor calibrated with a Hewlett-Packard 2804A quartz thermometer; the absolute cell temperature was always 25.00 f 0.02 OC. The densities of the pure components were also determined experimentally during each run. Measuring the pure component densities negates the possible error in vE due to small amounts (less than 1% ) of impurities in the pure components. This is true since any impurities would affect the mixture densities and pure component densities similarly, and vE is determined from the density differences between the mixture and the pure components. Materlals. Acetone, 2-butanone, carbon disulfide, carbon tetrachloride, chloroform, and ethyl acetate were Aldrich's HPLC grade, 99.9+%, and were used as purchased. Butylamine, furfural, heptane, hexane, methyl acetate, methyicyclohexane, 1-propanol, and vinyl acetate were Aldrich's, at least 99%, and were used as purchased. Ethanol was from U. S. Industrial Chemicals Co., 200 proof, and toluene was Baker Analyzed Reagent, stated assay 100%. Both were used as purchased. Nitroethane and nitromethane were Aldrich's, 96 % , and were fractionally distilled, the middle 60% being collected. I n all cases, nondegassed liquids were used. Data Reductlon The molar excess volumes were calculated from the density data by the equation
where x,, M,, p, are the mole fraction, molecular weight, and
0021-9568/87/ 1732-0083$01.50/0 0 1987 American Chemical Society
04 Journal of Chemical and Engineering Data, Vol. 32, No. 1, 1987
Table 11. Infinite Dilution P a r t i a l Molar Excess Volumes a t 25 "C
p,
DIE",
uO
0.01 002 0.03 0.04 0.05 MOLE FRACTION OF CHLOROFORM
Flgure 1. vE data for chloroform dilute in methyl acetate at 25 'C. The curve is the quadratic fit of the data.
density of component i , and pmis the density of the mixture. The partial molar excess volume of a component in a binary mixture can be determined from excess volume data V;E
=
VE
+ (1 - x;)(dvE/dx;)
component 1 acetone 2-butanone CHC13 acetone hexane heptane CS2 hexane CCl, heptane acetone ethanol methyl acetate ethyl acetate heptane nitroethane nitromethane
(3)
(4)
+ Bx,' x 1 x 2 / v E= A x , + B v E / x l = A + 6x1 vE = A x ,
(5) (6) (7)
Equations 4 and 5 simply represent linear and quadratic fits. Equation 6 is a van Laar type fit, which was used when the data displayed substantial curvature, as usually occurred in dilute alcohol solutions. Equation 7 is a fit of v E / x which is similar to the quadratic fit but has the effect of placing greater emphasis on the more dilute points. More complex fitting equations could be used, but since the curvature of the data is usually not great and since some scatter usually exists in the data, more complex expressions would not substantially improve the analysis. When suitable fiing equations had been applied to the data, the accuracy and relative merits of the correlations were examined and a SE" value was determined. An example of'such an analysis is shown in Table I for chloroform (1) dilute in methyl acetate (2). Since the data exhibit slight curvature, the linear and quadratic fitting equations were used. The quadratic
,,
cm3/mol 3.8 f 0.2 2.0 f 0.15 1.34 f 0.1 -0.20 f 0.1 8.0 f 0.5 4.7 f 0.25 6.6 f 0.35 5.1 f 0.4 4.2 f 0.3 0.60 f 0.1 -1.4 f 0.1 1.23 f 0.1 0.70 f 0.1 4.9 f 0.2 0.28 f 0.1 4.45 f 0.3
DIE"
system ethanol (1)in toluene (2) toluene (1)in ethanol (2) heptane (1) in ethanol (2) hexane (1) in ethanol (2) ethanol (1) in hexane (2)
At the limit of infinite dilution, vE and xi become zero
vE = Ax,
cm3/mol 5.0 f 0.3 3.3 f 0.2 2.2 f 0.2 -0.47 f 0.1 2.03 f 0.1 2.85 f 0.15 5.0 f 0.25 3.7 f 0.3 -0.75 f 0.1 1.9 f 0.15 -1.0 f 0.1 1.6 f 0.15 0.0 f 0.1 -0.42 f 0.1 -2.32 f 0.15 4.1 f 0.3 6.5 f 0.3
Table 111. Comparison of VIE" Values a t 25 "C Determined i n This Study to Literature Values
(2)
or the infinite dilution partial molar excess volume is the slope of the excess volume vs. mole fraction curve at infinite dilution. Thus, V,'" can be obtained by fiiing dilute v Evs. composition data to a curve, and obtaining the slope of the curve at xi = 0. Typically, five to ten data points at mole fractions ranging from approximately 0.005 to 0.05 were used to determine V/'-. Various equations were used to fit the data by the least-squares method. The equations employed depended on the amount of curvature of the v E data and on the number of data points taken. Since no theoretically correct equation exists, the uncertainties cited in the results reflect the minor differences in fitting with the various forms. The following fitting equations, where component 1 is the solute, were used:
component 2 hexane hexane hexane ethanol ethanol ethanol acetone vinyl acetate 1-propanol butylamine CHC13 toluene CHC13 CHCl, furfural hexane methylcyclohexane
this study 1.6 f 0.15 -1.4 f 0.1 2.85 f 0.15 2.03 f 0.1 8.0 f 0.5
lit. 1.495 -1.400 2.90 2.01 8.4
ref 4 4
3 3
5
equation gives a slightly better fit of the data, indicated by the smaller standard deviation (a). Thus the ViEm value for chloroform in methyl acetate is taken to be the A value for the quadratic equation, which is 1.23 cm3/mol. A graph of the v E data for this system is presented in Figure 1. Presentatlon of Results The 33 ViEm values obtained in this study are listed in Table The last three systems, heptane-furfural, nitroethanehexane, and nitromethane-methylcyclohexane, exhibit partial miscibility at 25 O C at higher concentrations than those studied here. The ViEm value of ethanol in heptane was found in the literature (3)and was not determined experimentally. The error estimates in Table I1 take into account the scatter of the v E data and uncertainty in Of"' due to substantial curvature of the data. Generally the results are accurate to fO.l cm3/mol or 5 8 % of ViE"', whichever is greater. An error analysis indicated that the vE values tend to be slightly less accurate for systems in which the solvent has a large molar volume (such as hexane or heptane), or in which the densities of the two components differ greatly. Conversely, greater accuracy in vE is obtained when the solvent has a small molar volume and when the two components have similar densities. In five cases, literature data were found that correspond to data obtained in this study. A comparison of these values is shown in Table 111. The agreement is quite satisfactory.
I I.
Acknowledgment The assistance of Paul Adriani and Brian Davison in performing the density measurements is sincerely appreciated. Reglslry No. CHCI,. 67-66-3; CS,, 75-15-0; CCI,, 56-23-5; methyl acetate, 79-20-9; acetone, 67-64-1; hexane, 110-54-3; 2-butanone, 7893-3; ethanol, 64-17-5; heptane, 142-82-5; vinyl acetate, 108-05-4; 1propanol, 7 1-23-8; butylamine, 109-73-9;toluene, 108-88-3;furfural, 980 1-1; nitroethane, 79-24-3; nitromethane, 75-52-5; methylcyclohexane. 108-87-2.
J. Chem. Eng. Data 1987, 32,85-92
Literature Clted
85
(4) Pardo, F.; Van Ness, H. C. J . Chem. Eng. Data 1965, 70, 163. (5) Marsh, K. N.; Burfitt, C. J . Chem. Thermodyn. 1975, 7 , 955.
(1) Kratky, 0.; Leopold, H.; Stabinger, H. Z . Angew. Phys. 1969, 2 7 , 273. (2) Sidiqi, M. A.; Gijtze, G.; Kohler, F. Ber. Bunsenges. Phys. Chem. 1980, 84, 529. (3) Van Ness, H. C.; Soczek, C. A.; Kochar, N. K. J . Chem. Eng. Data 1967, 12, 346.
Received for review May 5, 1986. Accepted July 21, 1986. J.H.C. received fellowships from the National Science Foundation and the €.I. DuPont de Nemours Co. The financial support of the E.I. DuPont de Nemours Co. is gratefully acknowledged.
Isopiestic Determination of the Osmotic and Activity Coefficients of Aqueous Mixtures of NaCl and MgCI, at 25 OC Joseph A. Rard” and Donald G. Miller University of California, Lawrence Livermore National Laboratory, Livermore, California 94550
The osmotic and activity coefficlents of aqueous mixtures of NaCl and MgCi, have been determined at 25 OC by uslng the isoplestic method. These measurements extend from low concentrations to the crystallization l i m b of the mixtures. They are critically compared to published Isopiestic, dlrect vapor pressure, and emf data for this system. Our data agree well with prevlous isoplestic data and two seis of emf values, but dlrect vapor pressure data are slgnlficantiy discrepant. Osmotlc and activity coefficients for NaCI-MgCI, mixtures are fairly reliably represented by both PHzer’s equations and Scatchard’s neutral electrolyte equations.
I ntroductlon A wide variety of chemical, geochemical, biochemical; and industrial systems and processes involve aqueous electrolyte mixtures. Osmotic and activity coefficient data are essential for these systems in order to quantitativelycharacterize and to model their reaction thermodynamics, their chemical speciation, and their solubiliiies. Activity data and their derivativesare also needed to calculate thermodynamic diffusion coefficients, which are based on chemical potential gradient rather than concentration gradient driving forces ( 7 -3). Several important ternary brine salt mixtures have been reinvestigated recently at 25 OC. Seidel et al. (4) and Kuschel and Seidel (5)have provided accurate data for aqueous KCIMgSO,, KCI-MgCI,, and K,SO,-MgSO, mixtures by using the isopiestic method. Ananthaswamy and Atkinson (6) studied NaCI-CaCI, mixtures by using Na+ and Ca2+ ion-sensitive electrodes, and Filippov and Cheremnykh (7) studied MgCI,MgS0, mixtures by using the isopiestic method. Another important brine salt mixture is NaCI-MgCI,. An extensive series of diffusion coefficient measurements is currently in progress in our laboratory. To analyze and interpret these data requires accurate activity coefficient derivatives. Since differentiation magnifies experimental errors, very precise experimental data are required. Several workers have previously studied aqueous NaCI-MgCI, at 25 OC. Experimental water activity data were reported from isopiestic measurements in two studies (8, 9 )and from direct water vapor measurements by the static method (70), and NaCl activities were reported in three studies ( 7 7 - 73)using emf measurements. Freezing point depression data ( 74) seem to be consistent with the 25 OC isopiestic values, but published enthalpy and heat capacity data are not adequate to convert 0021-956818711732-0085$01.50lO
the freezing point data to 25 OC. Thus, freezing point data will not be considered further. Additional enthalpy and heat capacity data would be desirable. The most extensive investigationof NaCI-MgCI, activities is Platford’s isopiestic study (9)which covers the ionic strength range of I = 0.11-8.02 mol-kg-l, where the kg refers to H,O. An et al.’s vapor pressure measurements (70) cover the ionic strength range of 6.37-13.84 mol-kg-’, but they are restricted to NaCl ionic strength fractions of 0.021-0.065. Osmotic coefficients from these two studies differ by 1-4% in the overlap region, and this exceeds the usual experimental errors for these methods by an order of magnitude. I n addition, both studies are somewhat scattered. The main source of these discrepancies seems to be the solution preparation method. Both studies (9,70) used direct weighing of “MgCI,.6H20” that had been dried with a water aspirator at 50 OC. I t has been our experience that attempts to dry hydrated chlorides to stoichiometric hydrates frequently results in samples that are overdried or underdried, so direct weighing of “MgCI,-6H,O” generally gives unreliable concentrations. Wu et al.’s isopiestic investigation (8) is free of this objection, but only 15 points were reported. Christenson (73) reported NaCl activities in NaCI-MgCI, at I = 1 mol-kg-’ and Lanier ( 7 7 ) at I = 1, 3, and 6 mobkg-l. Both used sodium-sensitive glass electrodes and Ag/AgCI electrodes. Butler and Huston (72)did a more detailed study, I = 0.52-5.99 mol-kg-‘, using Na(Hg)/Na+ and Ag/AgCI electrodes, but it was necessary to add 10.004 mol-kg-’ NaOH to their solutions in order to use the amalgam electrode. The data of Christenson (73)and Lanier ( 7 7 ) at I = 1 mol-kg-’ are in excellent agreement, with maximum differences of about 0.002 in the activity coefficients of NaCI. Butler and Huston ( 72) are in general agreement with the other two studies ( 7 7 , 73),but have about 5 times as much scatter. Unfortunately, at I = 6 mol-kg-‘ the only two studies ( 7 7 , 72)give inconsistent results. Lanier’s values are in general agreement with the isopiestic data. We, therefore, reject Butler and Huston’s data as inaccurate, most probably due to problems with the amalgam electrodes which are notoriously difficult to work with. As noted above, we need very accurate data for the calculation of activity derivatives. However, some of the available data for aqueous NaCI-MgCI, solutions are of doubtful quality, and the remaining data do not completely cover the accessible concentration regions. Consequently, we decided to reinvestigate this system by the isopiestic method. Our data extend from moderately low concentrations ( I = 0.295 mol-kg-’) to the
0 1987 American
Chemical Society